Bug 88598 – [Chromium-Android] Build DumpRenderTree with Android SDK

• Component: Tools / Tests
• Status: RESOLVED FIXED

[Xianzhu Wang]

This is follow-up of bug 86862, to change WebKit side accordingly after some changes at chromium side.

[Xianzhu Wang]

Created attachment 146468 [details]
patch

[Wei James (wistoch)]

*** Bug 88610 has been marked as a duplicate of this bug. ***

[Wei James (wistoch)]

*** Bug 88607 has been marked as a duplicate of this bug. ***

[Adam Barth]

Comment on attachment 146468 [details]
patch

I'm happy to rubber-stamp this change.  You might want to check with Peter before landing.

[Xianzhu Wang]

Comment on attachment 146468 [details]
patch

Peter (sorry I should have cc'ed you), does the patch look good to you? If yes, please just set cq+. Thanks.

We will still have many layout test failures with this patch.

Hao, could you create a meta bug for remaining Android DRT upstreaming tasks? Two things I just thought of are: test expectations, font settings.

[Peter Beverloo]

Comment on attachment 146468 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=146468&action=review

LGTM. I'm building the patch at the moment to verify it won't break the build. After it succeeds I'll put it on the commit queue. Cheers!

> Tools/DumpRenderTree/DumpRenderTree.gyp/DumpRenderTree.gyp:433
> +                    'action_name': 'apk_<(test_suite_name)',

Since this gyp-code is copied and not marked as FIXME, can we get rid of the test_suite_name variable and just fill in "DumpRenderTree"? At one point we'll move all of these support scripts (including generate_native_test.py, for example) to their own repository.

> Tools/DumpRenderTree/DumpRenderTree.gyp/DumpRenderTree.gyp:454
> +                        '--ant-compile',

I think this may need a rebase, given that the patch from bug 88626 is already on the commit queue.

> Tools/Scripts/webkitpy/layout_tests/port/chromium_android.py:276
> +        return self._build_path(configuration, 'DumpRenderTree_apk/DumpRenderTree-debug.apk')

Why do we use "-debug" here, again? Seems fine, but I forgot the reason :).

[Xianzhu Wang]

Comment on attachment 146468 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=146468&action=review

Will upload the new patch soon after I verify it locally.

>> Tools/DumpRenderTree/DumpRenderTree.gyp/DumpRenderTree.gyp:433
>> +                    'action_name': 'apk_<(test_suite_name)',
> 
> Since this gyp-code is copied and not marked as FIXME, can we get rid of the test_suite_name variable and just fill in "DumpRenderTree"? At one point we'll move all of these support scripts (including generate_native_test.py, for example) to their own repository.

Agreed. Done.

>> Tools/DumpRenderTree/DumpRenderTree.gyp/DumpRenderTree.gyp:454
>> +                        '--ant-compile',
> 
> I think this may need a rebase, given that the patch from bug 88626 is already on the commit queue.

Done.

>> Tools/Scripts/webkitpy/layout_tests/port/chromium_android.py:276
>> +        return self._build_path(configuration, 'DumpRenderTree_apk/DumpRenderTree-debug.apk')
> 
> Why do we use "-debug" here, again? Seems fine, but I forgot the reason :).

This is the naming convention of Android SDK's ant rule. In our ant build script (called from generate_native_test.py) the default type is 'debug'.

[Xianzhu Wang]

Created attachment 146624 [details]
patch for landing

Peter has verified it and admitted cq in offline.

[WebKit Review Bot]

Comment on attachment 146624 [details]
patch for landing

Rejecting attachment 146624 [details] from commit-queue.

Failed to run "['/mnt/git/webkit-commit-queue/Tools/Scripts/webkit-patch', '--status-host=queues.webkit.org', '-..." exit_code: 1

ERROR: /mnt/git/webkit-commit-queue/Tools/ChangeLog neither lists a valid reviewer nor contains the string "Unreviewed" or "Rubber stamp" (case insensitive).

Full output: http://queues.webkit.org/results/12924400

[Xianzhu Wang]

Created attachment 146633 [details]
patch for landing (with correct reviewer line)

[WebKit Review Bot]

Comment on attachment 146633 [details]
patch for landing (with correct reviewer line)

Clearing flags on attachment: 146633

Committed r119888: <http://trac.webkit.org/changeset/119888>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[Hao Zheng]

(In reply to comment #5)
> (From update of attachment 146468 [details])
> Peter (sorry I should have cc'ed you), does the patch look good to you? If yes, please just set cq+. Thanks.
> 
> We will still have many layout test failures with this patch.
> 
> Hao, could you create a meta bug for remaining Android DRT upstreaming tasks? Two things I just thought of are: test expectations, font settings.

Bug for font settings: https://bugs.webkit.org/show_bug.cgi?id=87006
We can file expectation bugs after we run layout test on some bot.